v0.5.0 — 2026 models, computer-use safety, NetOps pack

Zero-dependency linter for Claude Code / Agent Skills.

Highlights

• Current models: Opus 5, Sonnet 5, Fable 5, Mythos 5. Opus 4.8 / Sonnet 4.6 / Haiku 4.5 stay current.
• New checks: computer-unscoped, beta-header-stale, plugin-colon-name.
• NetOps pack: 22 production skills under skills/netops/.
• GitHub Action: uses: gesh75/claude-skill-lint@v0.5.0

Install

curl -fsSL https://raw.githubusercontent.com/gesh75/claude-skill-lint/main/install.sh -o install.sh
bash install.sh
skill-lint ~/.claude/skills

Docs: https://gesh75.github.io/claude-skill-lint/

v0.1.0 — first release

First public release of claude-skill-lint — a zero-dependency linter for Claude Code skills.

What it does

Audits a ~/.claude/skills directory and flags:

• broken / missing frontmatter (name, description)
• over-long descriptions (loaded into context on every turn)
• bloated bodies that should use progressive disclosure
• dead reference/ links
• stale Claude model IDs

Counts only real skills — a top-level <name>.md or a <dir>/SKILL.md.

Highlights

• Zero dependencies (Python 3.8+)
• JSON output, configurable thresholds, non-zero exit on errors (CI / pre-commit friendly)
• 9-test suite + GitHub Actions CI

Install

curl -O https://raw.githubusercontent.com/gesh75/claude-skill-lint/main/skill_lint.py
chmod +x skill_lint.py

MIT licensed.